Re: [PATCH v2 3/3] ci: stop linking built-ins to the dashed versions

From: Johannes Schindelin <johannes.schindelin@gmx.de>
Date: Aug 25, 2020, 08:30 UTC
Message-ID: <nycvar.QRO.7.76.6.2008251020260.56@tvgsbejvaqbjf.bet>
In-Reply-To: <xmqqimd750dr.fsf@gitster.c.googlers.com>

Hi Junio,

On Mon, 24 Aug 2020, Junio C Hamano wrote:

Show 21 quoted lines

> "Johannes Schindelin via GitGitGadget" <gitgitgadget@gmail.com>
> writes:
>
> > From: Johannes Schindelin <johannes.schindelin@gmx.de>
> >
> > Originally, all of Git's subcommands were implemented in their own
> > executable/script, using the naming scheme `git-<command-name>`. When
> > more and more functionality was turned into built-in commands (i.e. the
> > `git` executable could run them without spawning a separate process),
> > for backwards-compatibility, we hard-link the `git` executable to
> > `git-<built-in>` for every built-in.
> >
> > This backwards-compatibility was needed to support scripts that called
> > the dashed form, even if we deprecated that a _long_ time ago.
>
> This paragraph is irrelevant.  We are keeping the support for it and
> this topic is not newly deprecating or removing anything.  We need
> to argue for a need to test an installation that lacks these builtin
> subcommands anywhere on disk under their own names, which you did
> succinctly below (and there is no need for "For that reason,"
> there).

Could we please keep it? It will help me in the future when stumbling over this commit, to remember the context.

> > For that reason, we just introduced a Makefile knob to skip linking
> > them. TO make sure that this keeps working, teach the CI
>
> s/TO/To/

Thanks! I guess my keys got sticky or something ;-)

Show 7 quoted lines

> > (and PR) builds to skip generating those hard-links.
>
> What is not justified enough is why we no longer test installations
> with dashed builtins on disk.  If this topic is primarily about
> Windows (as 2/3 said), perhaps we can do this only for Windows tasks
> before we make a colletive decision to _DROP_ support for the on-disk
> builtin subcommands?

Oh, sorry, I will amend the commit message to clarify that the dashed form is actually not tested at all anymore. Specifically since e4597aae6590 (run test suite without dashed git-commands in PATH, 2009-12-02), in fact.

All this change does is to make it an even stronger committment to run the test suite without dashed Git commands.

Thanks, Dscho

Show 19 quoted lines

> > Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
> > ---
> >  ci/run-build-and-tests.sh | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/ci/run-build-and-tests.sh b/ci/run-build-and-tests.sh
> > index 6c27b886b8..1df9402c3b 100755
> > --- a/ci/run-build-and-tests.sh
> > +++ b/ci/run-build-and-tests.sh
> > @@ -10,7 +10,7 @@ windows*) cmd //c mklink //j t\\.prove "$(cygpath -aw "$cache_dir/.prove")";;
> >  *) ln -s "$cache_dir/.prove" t/.prove;;
> >  esac
> >
> > -make
> > +make SKIP_DASHED_BUILT_INS=YesPlease
> >  case "$jobname" in
> >  linux-gcc)
> >  	make test
>